Purposes: To improve normal tissue sparing for head-and-neck (H&N) image-guided radiotherapy
(IGRT) by employing treatment plans with tighter margins for CTV 2 and 3, and documenting the delivered
dose throughout the entire treatment course.
Methods: Ten H&N cases treated with simultaneous integrated boost on a TomoTherapy unit (Accuray Inc.)
were analyzed. Dose-limiting critical structures included brainstem, spinal cord, cochleae, parotid glands and
mandible. The targets include the PTV1 (gross disease volume), PTV2 (next echelon nodal regions) and
PTV3 (areas harboring subclinical disease). The standard margin plans (plan_ref) were generated using the
standard margin of 3 mm to CTV1-3. Reduced margin plans (plan_0margin) using the CTV-to-PTV margin
of zero for CTV2 and 3 were compared with plan_ref. All patients went through daily pre-treatment megavoltage
CT (MVCT) and weekly kilovoltage CT (kVCT) scans. A GPU-based 3D image deformation/
visualization tool was developed to register the weekly kVCT scans with the planning CT scan. The deformation
of each contoured structures was computed to account for non-rigid change in the patient setup.
Calculation of the dose accumulation was performed to determine the delivered mean/minimum/maximum
dose, dose volume histograms (DVHs), etc.
Results: The averaged planned cord maximum doses in Plan_0margin were 7.6% lower, and the parotid mean
doses were 18.9% lower than plan_Ref. No significant changes in D95 and D90 for the CTV2/3 cumulative
doses in both reference and Plan_0margin were observed during the planning stage. Under kVCT guidance on
TomoTherapy, for the reference plans, the averaged cumulative mean dose ratios during the entire treatment
course were consistent within 5% and 1.5% of the planned mean doses for PTVs and CTVs, respectively. Interfraction
anatomical changes introduced variations in delivered target doses that reduced the improved normal
structure sparing observed in plan_0margin during the planning stage. For the tighter margin plans, the
cumulative mean dose ratios were consistent within 4.3% and 2.3% of the planned mean doses for CTV2 and
CTV3, respectively. Similar dose variations of the delivered dose were seen for the reference and tighter margin
plans. However, the delivered maximum and mean doses for the cord were 20% and 10% higher than the
planned doses; a 3.6% higher cumulative mean dose for the parotids was also observed for the delivered dose
than the planned doses in both plans.
Conclusions: The GPU-based image framework enables real-time dose verification, accumulation and documentation.
By imposing tighter CTV margins for level 2 and 3 targets for H&N irradiation, acceptable cumulative
doses were achievable when coupled with weekly kVCT guidance while improving normal structure